pmxbot IRC Log Viewer

[06:59:47] <mgedmin> COOKIES!

[10:20:23] <mgedmin> why doesn't pip install .[test] work?

[14:34:36] <calston> I think there's a bug in pip freeze with git repos

[14:44:03] <calston> it's munging the #egg= portion with stuff it can't later parse

[14:47:50] <dstufft> calston: can you post steps to reproduce?

[14:48:03] <dstufft> mgedmin: the cookie thing is on my stack of things ot look at after pip 6 goes out fwiw

[14:48:18] <mgedmin> thank you!

[14:48:38] <calston> dstufft: https://gist.github.com/calston/e9b05b262f20865de9c0

[14:48:52] <dstufft> I'm really confused about what it could be, but I'm sure it's something related to PyPI being terrible

[14:48:56] <dstufft> calston: taking a look

[14:48:58] <calston> trying to figure out the code in pip but it's sort of spaghetti

[14:49:21] <dstufft> calston: is this using the latest develop?

[14:49:33] <calston> using whatever is the latest on pypi

[14:49:40] <dstufft> ok

[14:49:44] <dstufft> 1.5.6 then

[14:49:47] <calston> pip 1.5.6

[14:49:49] <calston> yes

[14:50:18] <dstufft> let me see if I can reproduce

[14:51:33] <dstufft> calston: I think I must be missing a step, I made a requirements.txt as you mentioned, then I did pip freeze and I didn't get an error

[14:52:48] <dstufft> oh I see

[14:52:53] <calston> the error is installing what pip freeze gives you

[14:52:54] <dstufft> you need to install the frozen

[14:52:57] <dstufft> yea

[14:52:59] <dstufft> ok I can repro

[14:53:03] <dstufft> one moment

[14:54:35] <dstufft> calston: so

[14:54:44] <dstufft> calston: good news, this is fixed in the devleopment version

[14:54:53] <dstufft> and I plan on releasing that today

[14:56:10] <calston> cool

[14:57:04] <calston> this haxy horrible way of distributing git repo requirements has never really worked anyway, but I got my hopes up

[16:22:16] <ionelmc> whoa new pip

[16:22:24] <ionelmc> any chance for new virtualenv? :)

[16:25:09] <Alex_Gaynor> ionelmc: Pretty sure it's coming

[16:27:58] <dstufft> yea it's coming

[16:30:38] <dstufft> waiting on https://github.com/pypa/virtualenv/pull/682

[16:37:54] <buck11> dstufft: is pip 6.0 stable, ready for prod?

[16:38:37] <dstufft> buck11: it's released yes

[16:38:49] <buck11> sweet

[16:38:58] <buck11> i imagine there's a sibling venv release

[16:39:29] <dstufft> there will be

[16:39:35] <dstufft> once the virtualenv tests pass

[16:40:14] <buck11> ah ok. i'll wait a bit then

[16:40:44] <dstufft> mostly the virtualenv tests are just waiting on the pip tests that got triggered in CI from me pushing things to pass

[16:40:52] <dstufft> from me pushing the verison bump*

[16:41:26] <buck11> how do tests wait on tests?

[16:41:56] <dstufft> because travis CI limits the number of concurrent builds per github org

[16:42:08] <buck11> dstufft: have you set sudo:false?

[16:42:27] <dstufft> and travis CI (mostly) runs them in order of what they are submitted

[16:42:32] <dstufft> buck11: I tried it, made things slower

[16:42:37] <buck11> sad!

[16:42:45] <dstufft> my theory is that the docker env doesn't have very good IO

[16:42:54] <dstufft> and pip's tests are super IO intensive

[16:43:14] <buck11> dstufft: about copying those fixtures, isn't that stuff readonly?

[16:43:24] <dstufft> if I run pip's tests with the same options on my iMac as travis runs them, my entire iMac grinds to a halt for awhile

[16:43:45] <buck11> the directories full of packages

[16:44:05] <dstufft> buck11: in theory yes, the problem is some of those fixtures are used by pip commands that modify the state of those directories

[16:44:16] <dstufft> realistically they should be split out into read only + not read only

[16:44:20] <dstufft> but I haven't had time

[16:44:28] <buck11> which pip commands do writes?

[16:44:37] <buck11> writes to the find-index area?

[16:44:52] <dstufft> pip install -e .

[16:44:53] <dstufft> etc

[16:44:59] <dstufft> they write egg_info directories

[16:45:02] <dstufft> to the local dir

[16:45:10] <buck11> ah i see

[16:45:48] <buck11> if the fixture instead set the thing chmod a-w, we'd be confident we'd caught all those

[16:46:18] <buck11> such a change would be a time investment for sure

[16:46:42] <ionelmc> dstufft: quite a shame, no setuptools upgrade?

[16:46:50] <dstufft> ionelmc: what do you mean

[16:47:11] <ionelmc> aaa

[16:47:12] <ionelmc> nevermind

[16:47:29] <ionelmc> seems 8.2 is in develop

[16:51:58] <dstufft> tdsmith: ping

[16:52:13] <buck11> dstufft: is it a simple matter to make a virtualenv package with pip 6.0 myself?

[16:52:34] <buck11> package of virtualenv*

[16:52:58] <dstufft> buck11: https://virtualenv.pypa.io/en/latest/virtualenv.html#the-extra-search-dir-option ?

[16:53:30] <buck11> dstufft: ah, yes

[16:58:33] <ionelmc> dstufft: circular imports in pip ? :(

[16:58:37] <ionelmc> https://www.irccloud.com/pastebin/hyxt2fQV

[17:00:07] <dstufft> ionelmc: hrm

[17:00:10] <dstufft> works fine here

[17:00:27] <dstufft> I wonder if it's because it's installed as an egg

[17:00:45] <ionelmc> it's definitely a circular import

[17:05:45] <dstufft> ionelmc: moment

[17:08:36] <ionelmc> dstufft: not trivial to solve cause there's lots of stuff in pip/__init__.py

[17:50:13] <ionelmc> dstufft: seems it only happens in the presence of /tmp/pip_build_foobar with wrong owner

[17:50:48] <dstufft> ionelmc: the circular import thing?

[17:50:54] <ionelmc> yeah

[17:52:31] <ionelmc> hmmmm

[17:52:54] <ionelmc> ah yes, cause there's no `pip.exceptions` at that point

[17:53:02] <ionelmc> it's still doing imports

[17:53:14] <tdsmith> dstufft: hello! ping for the pip release?

[17:54:37] <dstufft> tdsmith: Yea, though I can submit a PR, I just wanted to double check that my memory was right that you just wanted PRs against CPython for right now

[17:56:19] <tdsmith> don't worry about a PR; I want to make some picky changes to how python3 and the pypys anyway

[17:56:27] <dstufft> tdsmith: ok!

[17:57:14] <tdsmith> thanks for all your work on pip 6; glad to have it out!

[18:00:25] <ionelmc> dstufft: https://github.com/pypa/pip/pull/2247 should fix it

[18:00:30] <ionelmc> it was a bad error message

[18:00:39] <ionelmc> damn confusing :-)

[18:01:04] <dstufft> ionelmc: oh heh, I made https://github.com/pypa/pip/pull/2247 too lol

[18:01:24] <dstufft> tdsmith: np :)

[18:01:41] <ionelmc> dstufft: too slow, i won :-P

[18:01:46] <dstufft> er I mean https://github.com/pypa/pip/pull/2245

[18:01:52] <ionelmc> DAMN

[18:01:53] <ionelmc> nooo

[18:02:15] <ionelmc> actually you did different changes

[18:03:05] <tos9> wsgiref and argparse (for >py26) are now excluded from pip list and pip freeze

[18:03:07] <tos9> hooray?

[18:03:35] <ionelmc> dstufft: actually i think i fixed soemthing else

[18:04:07] <tos9> Use a randomized and secure default build directory when possible. <-- real hooray :)))))

[18:04:11] <dstufft> ionelmc: does my thing fix your problem?

[18:04:16] <tos9> dstufft: thanks, looks like a nice release.

[18:05:23] <ionelmc> dstufft: hmmm

[18:06:11] <ionelmc> dstufft: i think i may have misreported the issue :-|

[18:06:41] <ionelmc> dstufft: initially i had something this issue https://github.com/pypa/pip/issues/2067

[18:08:42] <dstufft> ionelmc: Ok, so I'm confused :D Is your pip currently broken with an error?

[18:10:14] <ionelmc> dstufft: ok, so it was an bad error message caused by a bad import (not a circular import) - pip would exit anyway at that point but that got me very confused (and I mistook everything for a circular import issue)

[18:10:29] <ionelmc> the fix is in https://github.com/pypa/pip/pull/2247

[18:10:41] <ionelmc> i'll add repro instructions there

[18:12:43] <Embo`work> What's up with pip 6.0? What happened to 2-5?

[18:13:01] <tomprince> Dropped the leading 1.

[18:13:18] <Embo`work> ah

[18:13:48] <ionelmc> Embo`work: it's like windows 10

[18:14:02] <dstufft> ionelmc: ok I see

[18:14:16] <ionelmc> dstufft: i've added details in the PR

[18:14:22] <Embo`work> It's breaking our jenkins builds for some reason, but that's just as much the fault of our outdated shiningpanda version which is still downloading it

[18:15:40] <Embo`work> for some reason pylint dies from missing deps. Hopefully we can get that jenkins plugin updated (though, why it has to have virtualenv plopped in it rather than using the system version, I really don't know)

[18:17:16] <dstufft> Embo`work: is it public? or can you paste a traceback

[18:17:16] <ionelmc> i have to say, the new pip has some nice stuff

[18:17:32] <ionelmc> is there some sort of highlevel overview with the new features?

[18:18:35] <Embo`work> dstufft: http://fpaste.org/162176/72202141/ << the relevant part

[18:19:18] <dstufft> Embo`work: hmm, that should work

[18:19:27] <dstufft> might be because of the 0.6c11 setuptools

[18:19:29] <dstufft> that's real old

[18:19:35] <dstufft> ionelmc: just the changelog really

[18:21:38] <Embo`work> I imagine it's that jenkins plugin. The dist-packages version of setuptools is 2.0.2, so I bet that old setuptools is probably coming from within the plugin

[18:25:33] <Embo`work> *sigh* >> https://github.com/jenkinsci/shiningpanda-plugin/blob/master/src/main/resources/jenkins/plugins/shiningpanda/workspace/setuptools-0-py2.py3-none-any.whl

[18:25:56] <Embo`work> I'm betting if I opened up that wheel, I'd find the old version inside.

[21:01:53] <AlexMeng> i'm assuming someone already mentioned this, but it looks like the newest release is "6.0" instead of "1.6"

[21:02:09] <AlexMeng> (for pip)

[21:04:44] <AlexMeng> or is the new version actually supposed to be 6.0?

[21:17:59] <Wooble> The 1. was dropped. On purpose.

[21:18:40] <Wooble> Just be glad it wasn't "14.0" or "2014.0"

[21:36:20] <buck11> I see virtualenv 12 just dropped. congrats! (no jinx)

[21:36:50] <buck11> AlexMeng: see the change notes

[21:50:44] <dstufft> Wooble: I wanted 14.0 ;)

[23:46:41] <_habnabit> heh how many people are going to come complain about the pip versioning

[23:47:58] <dstufft> _habnabit: there's been more than one

[23:48:08] <dstufft> I never realzied people noticed or cared

[23:48:33] <_habnabit> dstufft, i'm always surprised when people care about things

[23:57:23] <buck11> why is `session` a required argument to parse_arguments() now?

[23:57:27] <buck11> doesn't make sense to me

Log file Viewer

Help | Karma | Search:

#pypa logs for Monday the 22nd of December, 2014